"Global" top-level name proposed

[Dean Herington]

Simon Marlow wrote:

> Dean Herington writes:
> > The libraries description says about the "User" hierarchy:
> >
> > > The User hierarchy is reserved for users: a user may always use the
> > portion of the
> > > hierarchy which is formed from his/her email address as follows:
> > >     replace the @ by a ., reverse the order of the components,
> > capitalise the first letter of each >     component, and
> > prepend User..
> > For example,
> > >     simonmar@microsoft.com becomes User.Com.Microsoft.Simonmar.
> >
> > What if the username itself contains a '.'?  (This is fairly common.)
> > That could lead to ambiguity.  For example, (perhaps) both
> > "John.Smith@foo.com" and "John@smith.foo.com" translate to
> > "User.Com.Foo.Smith.John".
>
> Good point.  Replace dots in the username with underscores?

That would only trade one ambiguity for another.  Consider
"John.Smith@foo.com" and "John_Smith@foo.com" (unlikely, but possible).

How about a somewhat more far-reaching fix?  Recognizing that both "Org"
and "User" depend on the domain name hierarchy for their uniqueness, I
propose we replace "Org" and "User" by "Global" (in analogy with "Local").
An organization with domain name "foo.com" then has control over the
namespace beginning "Global.Com.Foo".  Such an organization is urged to
provide a sub-namespace "Global.Com.Foo.User" to provide uniqueness among
its users.  Note that this rule removes the possibility of ambiguity among
user names discussed in the earlier posts.  As before, one capitalizes (or
capitalises, if you're on the other side of the pond) the first letter of
each '.'-separated component of the user name.  Whether the '.'-separated
components should be reversed (as for the organization's domain name) is an
interesting choice; I slightly prefer reversing them, based on my
impression of the typical use of dotted user names.  Under that choice,
"john.smith@foo.com" would map to "Global.Com.Foo.User.Smith.John".

I note, for completeness, two assumptions that both the current proposal
and mine above make:
  * It is assumed that the case of the initial letters of '.'-separated
components (of both domain names and user names) does not matter for
uniqueness.
  * It is assumed that the authority that controls the domain namespace
with '.'-separated components [C(n) .. C(1)], where n>0, also controls the
namespace with components [X, C(n) .. C(1)], for all X.  The controlling
authority can, of course, pass on or delegate authority for
sub-namespaces.  For example, it is assumed that "smith.foo.com" is under
the umbrella of "foo.com".  Without this assumption, there could be two
authorities for the "Global.Com.Foo.Smith" namespace.

Dean